基于异质传感器信息融合的移动机器人同步定位与构图

摘要

Aiming at the problem of low locationing accuracy and incomplete mapping in simultaneous localization and mapping (SLAM) using single sensor of mobile robot,a SLAM algorithm based on the information fusion of Kinect vision sensor and laser sensor was proposed.Firstly,the depth image from Kinect sensor are converted to a 3 D point cloud through the coordinate system conversion,and the height information of the 3 D point cloud is filtered by vertical limit filter.Then the remaining 3D point cloud is projected to the horizontal plane and the boundary point cloud information are extracted to convert into the laser scan data.After that the converted laser scan data fused with the scan data from laser sensor at the data-level.Finally,unified data was output to realizing the mapping and autonomous navigation of mobile robot.The experimental results show that this method can accurately detect small and complex features obstacles,create accurately and fully environment map,and also better complete the task of the mobile robot autonomous navigation.%针对采用单一传感器在移动机器人同步定位与构图(SLAM)中存在定位精度低、构图不完整等问题,提出一种基于Kinect视觉传感器和激光传感器信息融合的SLAM算法.首先将Kinect传感器获取的深度图像,经过坐标系转换得到三维点云、通过限制垂直方向滤波器过滤三维点云的高度信息;再将剩余三维点云投影到水平面并提取边界点云信息转化为激光扫描数据;然后与激光传感器的扫描数据进行数据级的信息融合;最后输出统一数据实现移动机器人的构图及自主导航.实验结果表明,该方法能够准确地检测小的,及特征复杂的障碍物,能够构建更精确、更完整的环境地图;且更好地完成移动机器人自主导航任务.